PDCAT: a framework for fast, robust, and occlusion resilient fiducial marker tracking
- PDF / 2,880,480 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 46 Downloads / 211 Views
ORIGINAL RESEARCH PAPER
PDCAT: a framework for fast, robust, and occlusion resilient fiducial marker tracking Oualid Aaaar1 · Imad Eddine Mokhtari1 · Mohamed Bengherabi1 Received: 3 May 2020 / Accepted: 13 August 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Square binary patterns have become the de facto fiducial marker for most computer vision applications. Existing tracking solutions suffer a number of limitations, such as the low frame-rate and sensitivity to partial occlusions. This work aims at overcoming these limitations, by exploiting temporal information in video-sequences. We propose a parallel detection, compensation and tracking (PDCAT) framework, which can be integrated into any binary marker system. Our solution is capable of recovering markers even when they become mostly occluded. Furthermore, the low processing time of the tracking task makes PDCAT more than an order of magnitude faster than a track-by-detect solution. This is particularly important for embedded computer vision applications, wherein the detection run at a very low frame rate. In the experiments conducted on an embedded computer, the processing frame rate of the track-by-detect solution was merely 11 FPS. Our solution, on the other hand, was capable of processing more than 100 FPS. Keywords Fiducial marker · Detection · Tracking · Embedded · Real-time · AprilTag
1 Introduction Pose estimation is a fundamental problem in many computer vision applications, including robot navigation [1, 2], visual servoing [3, 4], augmented and virtual reality [5, 6]. The pose of a monocular camera can not be recovered unless a minimum set of correspondences between the 3D world and 2D image space is determined. This can be achieved using natural features extracted online and matched against a 3D model of the scene. The 3D model can be obtained from CAD files or built offline using techniques such as structure from motion (SFM), or online using simultaneous localisation and mapping (SLAM) [7]. The advantage of using natural features is that no modification to the environment is required. The performance of such techniques, however, significantly degrades in poorly textured scenes. In this case, the usage of an active or passive fiducial marker becomes indispensable [8]. Active markers are designed using light emitting sources, commonly diodes, which can be easily segmented from the * Oualid Aaaar [email protected]; [email protected] 1
Ecole Militaire Polytechnique, Bordj el Bahri, B.P 17, Algiers, Algeria
background. A problem with this solution is that the pose of an object cannot be obtained unless several markers are combined. This requires solving an assignment problem for identifying each marker. Another drawback is the limited number of patterns which can be achieved. Passive markers can be designed based on colour information [9, 10], by building color patterns in such a way to make them distinguishable from the background. The usage of colour information permits a fast detection and identificat
Data Loading...